WEBVTT - Meta Turns to Nuclear Power to Run AI 0:00:00.800 --> 0:00:05.120 From the heart of where innovation, money and power collide 0:00:05.320 --> 0:00:10.600 in Silicon Valley and beyond. This is Bloomberg Technology with 0:00:10.760 --> 0:00:12.240 Caroline Hyde. 0:00:11.840 --> 0:00:13.320 And Ed Ludlow. 0:00:26.760 --> 0:00:28.960 Now from New York and San Francisco. This is Bloomberg 0:00:29.000 --> 0:00:32.960 Technology coming up. Meta goes nuclear and buys nuclear power 0:00:32.960 --> 0:00:35.800 from Constellation Energy for its AI needs. 0:00:36.520 --> 0:00:39.400 Plus Elon Musk selling five billion dollars in debt for 0:00:39.440 --> 0:00:43.600 his own AI startup, Xai, with Morgan Stanley shopping that debt. 0:00:43.640 --> 0:00:46.839 And we sit down with Drew Baglino, former Tesla executive, 0:00:46.960 --> 0:00:49.280 who's out with his new startup, Heron. 0:00:49.680 --> 0:00:50.680 But first we check in on. 0:00:50.680 --> 0:00:52.960 These markets and look energy dictating trade in many ways 0:00:52.960 --> 0:00:54.760 in the tech news, but there's energy in some of 0:00:54.760 --> 0:00:55.320 the benchmarks. 0:00:55.320 --> 0:00:58.600 We're looking at NASDAC now up six ten percent ed. 0:00:58.920 --> 0:01:01.520 This is as we see the job stub must coming in, 0:01:01.720 --> 0:01:05.640 showing resilience for US economic growth, even though the OECD 0:01:05.880 --> 0:01:08.680 had pointed to some anxiety around growth and the tariff 0:01:08.680 --> 0:01:09.800 impact earlier today. 0:01:10.000 --> 0:01:12.199 But you're looking under the hood. What's caught your attention? 0:01:13.120 --> 0:01:16.840 Yeah, that Meta nuclear deal. Meta shares slightly softer Constellation 0:01:17.319 --> 0:01:21.759 supplying the energy. The nuclear energy actually really muted gains 0:01:21.800 --> 0:01:23.479 relative to the big jump we saw in the pre 0:01:23.520 --> 0:01:26.600 market when the deal was announced one and twenty one 0:01:26.640 --> 0:01:30.520 megawatts over twenty years. This isn't just a sizeable deal 0:01:30.760 --> 0:01:33.320 to source nuclear energy for Meta in its data centers. 0:01:33.560 --> 0:01:36.959 It's bigger than the deal that Microsoft did with Constellation, 0:01:37.040 --> 0:01:38.839 and it's worth getting into the details on this. 0:01:38.760 --> 0:01:40.800 One character and we can we got the perfect person 0:01:40.800 --> 0:01:42.520 the Blueblegs will weighed is with us. 0:01:42.760 --> 0:01:44.360 The context here is important. 0:01:44.760 --> 0:01:47.960 It is bigger than the Microsoft purchase that we'd seen previously, 0:01:48.440 --> 0:01:51.000 and what does it allow Meta to do in terms 0:01:51.000 --> 0:01:54.800 of yet further energy coming from a for AI from 0:01:54.840 --> 0:01:55.920 nuclear in the future too. 0:01:56.240 --> 0:01:58.520 Well, the thing about nuclear is that it's clean. There's 0:01:58.560 --> 0:02:01.960 no carbon from nuclear power plants and wind and solar 0:02:02.000 --> 0:02:04.440 or clean and all these tech companies have these really 0:02:04.440 --> 0:02:07.240 ambitious carbon free goals. We're going to be in that 0:02:07.360 --> 0:02:11.679 zero by twenty thirty, thirty five whatever. In the last year, 0:02:11.760 --> 0:02:14.840 their power needs have climbed and they need so much 0:02:14.880 --> 0:02:17.880 electricity for all the SAYI power plants take a long 0:02:17.919 --> 0:02:20.640 time to build, so suddenly they're like turning back to 0:02:20.720 --> 0:02:24.320 gas and coal, and that's a threat to their green goals. 0:02:24.400 --> 0:02:26.799 So being able to get nuclear that's a real win. 0:02:28.400 --> 0:02:31.120 Well on the Constellation side, there's a lot of signals 0:02:31.160 --> 0:02:37.639 here of commitment upgrading and updating capacity, potentially building new capacity. 0:02:37.919 --> 0:02:40.200 What have you learned in your reporting about this company. 0:02:40.639 --> 0:02:44.040 Yeah, so that's really exciting because we don't have any 0:02:44.280 --> 0:02:47.760 big nuclear plants under construction right now. The last one 0:02:48.080 --> 0:02:51.000 was the Vocal plant to Georgia, and they finished. 0:02:51.000 --> 0:02:52.200 It was kind of a debacle. 0:02:52.200 --> 0:02:55.440 It was seven years late, it was billions over budget, 0:02:55.440 --> 0:02:57.720 and it really scared people away from wanting to do 0:02:57.760 --> 0:03:00.440 that again. Even though the rest of the country you're saying, 0:03:00.440 --> 0:03:03.519 please build us more big nuclear power plants. The only 0:03:03.560 --> 0:03:06.760 people that aren't saying that are companies like Constellation that 0:03:06.960 --> 0:03:08.880 have to write the checks for these power plants. 0:03:09.200 --> 0:03:10.200 So if they can. 0:03:10.120 --> 0:03:13.880 Line up a customer like Microsoft, like Meta, someone who's 0:03:13.919 --> 0:03:17.160 going to be willing to help them out on the finances, 0:03:17.480 --> 0:03:21.840 that's making them have more confidence about building a new plant. 0:03:21.919 --> 0:03:26.640 And Constellation said exactly that they have an NRC permit 0:03:26.720 --> 0:03:29.480 that gives them permission to do another reactor at the 0:03:29.480 --> 0:03:32.120 same Clinton side in Illinois. So there's not very many 0:03:32.160 --> 0:03:35.520 places that already have that paper. We're handled and so 0:03:35.840 --> 0:03:38.840 Constellation CEO Joe Dimingas told me last week that yeah, 0:03:38.840 --> 0:03:40.520 they're looking at it. They're looking at a lot of 0:03:40.560 --> 0:03:43.960 different options, but they're seriously considering it, and that's something 0:03:44.000 --> 0:03:45.000 we haven't heard in the while. 0:03:45.960 --> 0:03:48.720 Blim bogs will wade, Thank you very much. Let's now 0:03:48.760 --> 0:03:50.800 expand the conversations, take a look at the impact of 0:03:50.840 --> 0:03:54.720 the deal for both the technology and nuculiar power industries. 0:03:54.760 --> 0:03:57.720 Joining us as Jeffrey's analyst Brent Hill and I've done 0:03:57.720 --> 0:04:00.880 the math on it right. Basically, this deal between Constellation 0:04:00.920 --> 0:04:04.640 and Meta is about thirty five percent larger in total 0:04:04.720 --> 0:04:08.680 energy terms on constant capacity basis than the Microsoft deal. 0:04:08.920 --> 0:04:12.720 But it's worth putting Microsoft as a Meta as an 0:04:12.720 --> 0:04:16.720 operator of data centers into context versus the hyperscalers. What 0:04:16.800 --> 0:04:19.760 does that math tell you about Meta and what it's 0:04:19.760 --> 0:04:20.320 doing here. 0:04:21.760 --> 0:04:24.119 Well, Meta's playing big for the AI game. 0:04:24.440 --> 0:04:27.599 They're going to spend over sixty five billion dollars this 0:04:27.720 --> 0:04:31.400 year in CAPEX and they have one of the highest 0:04:31.400 --> 0:04:35.839 increases in terms of absolute dollars spent on CAPEX to 0:04:35.920 --> 0:04:40.880 fuel this AI generation. So they're a huge player along 0:04:40.880 --> 0:04:45.479 with the other hyperscalers. Amazon in microsoftware obviously going to 0:04:45.480 --> 0:04:47.560 spend more in capex, but the percent. 0:04:47.480 --> 0:04:50.080 Increases the highest at MATA. 0:04:50.880 --> 0:04:53.080 And I think what this deal signifies is that we're 0:04:53.120 --> 0:04:55.920 at the beginning of this aid option curve. It's going 0:04:55.960 --> 0:04:59.160 to take we're again any one if you will, have 0:04:59.200 --> 0:05:02.560 a baseball on l UH and this is going to 0:05:02.600 --> 0:05:07.600 be a multi deck guy decade transformation. I think when 0:05:07.640 --> 0:05:11.239 you look at the statistics of this transaction, right it's 0:05:11.360 --> 0:05:13.480 a twenty year deal, which puts it at the higher 0:05:13.560 --> 0:05:17.320 end of of of an energy transaction. They're pain actually 0:05:17.320 --> 0:05:20.719 more than they would for other energy. They're not building 0:05:20.800 --> 0:05:23.680 the data center near the nuclear plant. It's all virtual, 0:05:24.120 --> 0:05:27.080 so I think they're effectively as as yours your other 0:05:27.120 --> 0:05:30.120 presenter mentioned, this is an offset from a from a 0:05:30.160 --> 0:05:35.040 carbon perspective, so better for the environment. And I think 0:05:35.040 --> 0:05:38.080 they obviously want to get as much a capacity locked 0:05:38.120 --> 0:05:40.040 up as they can before others get. 0:05:39.839 --> 0:05:42.839 To that to get get to those uh that capacity 0:05:42.839 --> 0:05:43.400 and energy. 0:05:43.480 --> 0:05:45.880 So I think, I mean, there's multiple things to read 0:05:45.920 --> 0:05:48.280 in here and look at from from our side. We've 0:05:48.279 --> 0:05:51.080 talked to our energy team and and and this again 0:05:51.600 --> 0:05:54.280 goes to the longer end and longer tail. And obviously 0:05:54.320 --> 0:05:57.600 they wanted to save this facility, so this obviously gives 0:05:57.640 --> 0:06:00.240 them a chance to to basically take a facility that 0:06:00.279 --> 0:06:02.760 maybe was going to be shut down to saving it. 0:06:02.960 --> 0:06:06.600 Brent, how much anxiety have you had about energy being 0:06:06.640 --> 0:06:10.279 the blocker here to these hyperscale is meeting at the 0:06:10.360 --> 0:06:12.039 level of demand massive. 0:06:12.200 --> 0:06:14.279 I mean, I'm a tech guy, and again I was 0:06:14.320 --> 0:06:16.200 just talking to our energy team before I got on. 0:06:16.240 --> 0:06:17.040 I mean, who would have. 0:06:16.960 --> 0:06:21.560 Thought that at this point we'd all be scrambling to 0:06:21.640 --> 0:06:23.760 understand energy real estate. 0:06:24.000 --> 0:06:24.400 I mean, the. 0:06:24.320 --> 0:06:29.400 Biggest bottleneck to powering AI right now are the transmission 0:06:29.480 --> 0:06:33.760 lines from the energy companies to the data centers, and 0:06:33.839 --> 0:06:36.359 these you know, we've been told by many that are 0:06:37.000 --> 0:06:40.320 sometimes two or three years backlogged, so they can't even 0:06:40.560 --> 0:06:42.960 they can get things built, but they can't actually finish 0:06:43.080 --> 0:06:46.640 the last piece of this because their lead times are 0:06:46.680 --> 0:06:50.160 so long to get the equipment. So, I mean, we're 0:06:50.360 --> 0:06:52.760 again Microsoft has said this to us many times. 0:06:52.960 --> 0:06:53.880 When they show up. 0:06:53.760 --> 0:06:57.760 Every day, they're like it is like they're on fire 0:06:57.839 --> 0:07:00.440 the whole day, where they just can't like sit because 0:07:00.480 --> 0:07:03.599 there's so many mooky pieces to make this all work. 0:07:04.520 --> 0:07:07.640 For us to basically hit on one prompt bar and say, 0:07:08.560 --> 0:07:11.600 you know, ask a simple question, what hotel should I 0:07:11.600 --> 0:07:12.640 stay at in. 0:07:12.600 --> 0:07:15.080 San Francisco based on this budget? 0:07:15.160 --> 0:07:15.320 You know? 0:07:15.440 --> 0:07:17.200 Or whatever question you want to ask. 0:07:17.240 --> 0:07:19.200 But it's like you think about how simple the questions are, 0:07:19.520 --> 0:07:21.360 but what actually has to happen in the back end 0:07:21.440 --> 0:07:25.880 is so complex, and so we're living with massive anxiety 0:07:25.880 --> 0:07:27.960 even covering the space, trying to keep up with all the. 0:07:28.000 --> 0:07:30.360 Data pa Isn't it funny that it feels but a 0:07:30.360 --> 0:07:32.760 few weeks ago that the anxiety was that they were 0:07:32.800 --> 0:07:36.679 overbuilding and that the supply side was just getting too much. 0:07:36.800 --> 0:07:38.040 Is that a thing of the past. 0:07:37.800 --> 0:07:40.720 Brent, It's a thing of the past. I think here's 0:07:40.760 --> 0:07:41.080 the problem. 0:07:41.120 --> 0:07:45.320 We're doing third grade math and nothing against my competitors, 0:07:45.520 --> 0:07:47.920 but all my competitors put out a data point though, Hi, 0:07:48.000 --> 0:07:52.360 a data center for Microsoft is on hold, the Wisconsin 0:07:52.640 --> 0:07:55.520 facilities not coming up live, and everyone goes, hey, i's 0:07:55.600 --> 0:08:00.200 over everyone, run for the hills, sell everything, and that's out. 0:08:00.200 --> 0:08:00.760 What's happening. 0:08:00.760 --> 0:08:04.080 What's happening is we're doing this efficiently and we're going 0:08:04.120 --> 0:08:07.200 to move pieces around. But as Microsoft and others have shown, 0:08:07.440 --> 0:08:11.400 they're making money h and they're actually generating revenue growth. 0:08:11.560 --> 0:08:14.679 So it's not about just blowing money and not generating 0:08:14.720 --> 0:08:18.640 a return. It's it's it's balanced revenue growth and profitability. 0:08:19.000 --> 0:08:21.480 And I think many of the companies that we cover 0:08:21.920 --> 0:08:23.920 are signaling that they're going to be pieces that move 0:08:23.960 --> 0:08:26.600 all over and sometimes again, as I said, you can't 0:08:26.600 --> 0:08:28.160 even find the transmission lines to. 0:08:28.120 --> 0:08:30.880 Get to these dcs, so they have to move a 0:08:30.960 --> 0:08:35.839 capacity to other data centers. And so it's yeah, it's 0:08:35.840 --> 0:08:39.520 a it's a giant rubics cube that is super scrambled, 0:08:39.559 --> 0:08:41.080 and I'm not very good at a rubics cubes. 0:08:41.120 --> 0:08:43.400 So we're trying to we're trying to figure it out. 0:08:44.120 --> 0:08:46.400 Brent, neither am I to be fair, But but what 0:08:46.440 --> 0:08:48.719 you just said it makes me want to go back 0:08:48.720 --> 0:08:52.280 to the twenty year commitment with this particular nuclear deal, 0:08:52.679 --> 0:08:56.160 because you can either see it as absolute conviction of 0:08:56.200 --> 0:08:59.360 the need to source energy at those levels long term 0:08:59.440 --> 0:09:02.200 because of what AI is doing, or you look at 0:09:02.200 --> 0:09:04.880 it as a hedge the need for protection to make 0:09:04.920 --> 0:09:07.600 sure you've got the energy security if you need it. 0:09:09.440 --> 0:09:10.880 I mean, I think it's a combo of both. 0:09:10.920 --> 0:09:13.600 And again it's it's a carbon offset right from all 0:09:13.600 --> 0:09:15.360 the other ways that they're generating power. 0:09:15.400 --> 0:09:16.360 So there's there's. 0:09:16.160 --> 0:09:19.840 That, you know, environmental component that they have to be 0:09:19.960 --> 0:09:22.720 conscious of. So I think they check the box on 0:09:22.800 --> 0:09:25.680 that they lock up a facility that perhaps was going 0:09:25.760 --> 0:09:27.040 to go offline. 0:09:26.760 --> 0:09:28.079 They protect this. 0:09:28.200 --> 0:09:31.800 Power from perhaps others that we're going to get it 0:09:32.440 --> 0:09:35.599 in the AI industry. Uh, you know, there's kind of 0:09:35.640 --> 0:09:38.000 multiple you know, checkboxes. 0:09:38.040 --> 0:09:39.800 I think they get they get out of this transaction 0:09:40.240 --> 0:09:41.960 as we understand it. It's virtual. 0:09:42.240 --> 0:09:45.200 There's no like Amazon and others are are doing nuclear 0:09:45.200 --> 0:09:48.319 and putting DCS near it. As we understand that this 0:09:48.440 --> 0:09:51.600 is nothing's near it. So you know, it's a virtual deal, 0:09:52.040 --> 0:09:56.320 which again is unique. So I think there's kind of 0:09:56.480 --> 0:10:01.280 multiple threads to pull here. But again I I think 0:10:01.440 --> 0:10:05.360 what this signifies is that we are really early. It's 0:10:05.400 --> 0:10:08.520 going to have massive impacts on all of our lives 0:10:08.840 --> 0:10:12.080 across every industry, and I think they talked about this yesterday. 0:10:12.480 --> 0:10:14.840 Over the next year, Meta will be able to power 0:10:15.000 --> 0:10:20.520 full advertising AI actual campaigns, which means if you run 0:10:20.520 --> 0:10:22.600 a small business, you can say I'd like a sidewalk 0:10:22.640 --> 0:10:26.400 sale for my furniture company, and here are the pictures 0:10:26.400 --> 0:10:29.120 from the last year. I don't have to hire a videographer. 0:10:29.160 --> 0:10:31.760 I don't have to hire a creative pro whatever. You 0:10:31.800 --> 0:10:33.360 just put the pictures in. All of a sudden, it 0:10:33.400 --> 0:10:37.200 spits out, here's a sidewalk sale Thursday with the band, 0:10:37.920 --> 0:10:41.400 the wine, the cheese, and here's the discounts, and here's 0:10:41.440 --> 0:10:43.120 the video, and you do it at a fraction of 0:10:43.160 --> 0:10:43.440 the costs. 0:10:43.480 --> 0:10:45.240 I mean, this is pretty incredible what this is going 0:10:45.280 --> 0:10:46.959 to do for small businesses. 0:10:46.800 --> 0:10:51.120 And metas is enabling this small business economy to define 0:10:51.120 --> 0:10:55.120 the right customers, to generate the right campaigns. 0:10:54.120 --> 0:10:58.040 It's it's a completely different world. And again, I think 0:10:57.880 --> 0:11:00.280 we're all trying to have trying to. 0:11:00.120 --> 0:11:02.040 We're trying to have more wine and cheese with our 0:11:02.120 --> 0:11:05.079 generator AI. Brent, that's what we're trying to do. Jeffrey's analyst, 0:11:05.440 --> 0:11:07.600 Brent phil is so good to have you on just 0:11:07.600 --> 0:11:09.800 the scale of the ambition coming from Meta. Meanwhile, let's 0:11:09.800 --> 0:11:12.160 talk about what's happening with Morgan. Stanley is shopping a 0:11:12.200 --> 0:11:15.560 five billion dollar debt package for Elon Musk's AI company Xai. 0:11:15.720 --> 0:11:16.760 It's according to a source. 0:11:16.920 --> 0:11:19.079 Now, this is the latest fundraising move by the billionaire, 0:11:19.080 --> 0:11:21.360 who is also tapping markets of fresh equity across his 0:11:21.400 --> 0:11:24.680 Business Empire and most Kurt Wagner joins us some more. Look, 0:11:24.760 --> 0:11:27.200 is he looking to the debt market to finance his 0:11:27.200 --> 0:11:28.360 own infrastructure build out? 0:11:29.920 --> 0:11:32.960 I have to imagine so, because if you remember Caroline, 0:11:33.080 --> 0:11:36.280 XAI has that huge data center in Memphis, which they 0:11:36.360 --> 0:11:38.880 call Colossus. It was just I think last month that 0:11:38.920 --> 0:11:41.120 Elon Musk was out saying that, you know, they have 0:11:41.160 --> 0:11:44.160 two hundred thousand GPUs at that site, with aims to 0:11:44.200 --> 0:11:47.560 be more than a million GPUs either there or even 0:11:47.640 --> 0:11:50.440 at a site nearby. So I mean the scale at 0:11:50.440 --> 0:11:53.559 which Elon is thinking as it comes to XAI and 0:11:53.880 --> 0:11:56.720 these data centers is very massive. Obviously, as we've talked 0:11:56.720 --> 0:11:58.880 about and you know as well as anybody, this is 0:11:58.920 --> 0:12:02.079 incredibly expensive. So it's not surprising to me that he's 0:12:02.120 --> 0:12:04.240 not only out trying to raise this money via debt, 0:12:04.280 --> 0:12:06.839 but also, as Bloomberg has reported, you know, they're looking 0:12:06.880 --> 0:12:09.720 to raise money with an equity round as well. 0:12:11.200 --> 0:12:13.679 This is a lot of things happening in parallel. You 0:12:13.720 --> 0:12:17.520 have debt. What was reported yesterday by another outlet is 0:12:17.559 --> 0:12:21.760 a tender to give employees liquidity both XAI and ex 0:12:21.800 --> 0:12:24.800 employees a secondary offering and then go back to that 0:12:24.840 --> 0:12:27.559 bit Kert. What I'd heard was that as soon as 0:12:27.600 --> 0:12:31.240 that merger closed, Elon and people like Jared Birchill, his 0:12:31.440 --> 0:12:35.000 financial lieutenant, side phoning everyone and saying, if we raise 0:12:35.040 --> 0:12:38.160 some money, want to get involved. That's kind of classic 0:12:38.520 --> 0:12:42.240 Elon Inc. Play, it is, and this. 0:12:42.160 --> 0:12:43.840 Is what Elon does, right. 0:12:43.920 --> 0:12:47.200 He spends a lot of his time over in government, 0:12:47.320 --> 0:12:51.800 sort of seemingly like ruining his reputation and watching Tesla's 0:12:51.800 --> 0:12:54.800 stock tank, and then at the same time people are 0:12:54.880 --> 0:12:57.200 lining up out the door to do business with him 0:12:57.320 --> 0:13:00.560 in all of these private endeavors. As you may right, 0:13:00.880 --> 0:13:04.760 XAI Neuralink just raised over six hundred million dollars. 0:13:04.800 --> 0:13:06.240 I believe they announced that yesterday. 0:13:06.920 --> 0:13:09.280 So you know, this is someone who, because of his 0:13:09.440 --> 0:13:13.160 track record, because of his ambitions, usually has no problem 0:13:13.200 --> 0:13:16.120 getting people to give him money. I think the question 0:13:16.280 --> 0:13:19.160 is how much and where does it go right, How 0:13:19.800 --> 0:13:22.720 useful is it for him or is he just raising 0:13:22.760 --> 0:13:24.600 money for the sake of raising money. I think in 0:13:24.640 --> 0:13:27.280 this case with the XAI thing, given what he signaled 0:13:27.280 --> 0:13:30.080 publicly about the data centers and Memphis specifically, I think 0:13:30.080 --> 0:13:32.160 we can assume a lot of this would go toward 0:13:32.200 --> 0:13:34.959 building out that project. 0:13:34.320 --> 0:13:36.800 Bloom bes cut Wagon that thank you very much. Coming up, 0:13:36.840 --> 0:13:40.280 an energy startup founded by a former Tesler, ezech is 0:13:40.320 --> 0:13:44.920 aiming to transform transformers. We'll talk to the see Drew 0:13:44.920 --> 0:13:48.480 Baglino about the company's latest fundraise and his technology. Next, 0:13:48.840 --> 0:14:03.280 this is Bloomberg Technology. Increase the electricity demand driven by 0:14:03.320 --> 0:14:07.400 AI is running up a shortage of transformers. Startup Heron 0:14:07.520 --> 0:14:10.320 Power has set out to address that bottleneck with solid 0:14:10.360 --> 0:14:14.120 state transformers that help electricity in its sector scale faster. 0:14:14.200 --> 0:14:17.920 Its founder, Drew Baglino, is the high profile, long serving 0:14:18.160 --> 0:14:21.880 and now former Tesla exec who helps shape that company's powertrain, 0:14:21.960 --> 0:14:25.120 battery and energy product strategy. Delight to say, Drew joins 0:14:25.200 --> 0:14:26.640 us here in San Francisco. 0:14:27.000 --> 0:14:27.880 Thanks for having me ed. 0:14:27.960 --> 0:14:29.880 A lot of people were waiting to see what you 0:14:29.880 --> 0:14:32.520 would do in your life after Tesla. It is Heron 0:14:32.680 --> 0:14:37.160 and Heron Power. What is the technology and what's the goal. Yeah. 0:14:37.280 --> 0:14:42.160 Heron Power's goal is to build grid scale power solutions 0:14:42.200 --> 0:14:46.080 to an accelerate and all electric future. The technology is 0:14:46.120 --> 0:14:50.680 basically leveraging the bleeding edge of power semiconductors to go 0:14:50.800 --> 0:14:54.120 directly from all the new energy technologies that are out there. 0:14:54.120 --> 0:14:55.000 They're all DC. 0:14:55.240 --> 0:14:58.400 This is solar batteries, data centers, and they got to 0:14:58.440 --> 0:15:01.280 get to a grid which is ac alter current. And 0:15:01.560 --> 0:15:05.320 we're going to do that without transformers, fully with power electronics, 0:15:05.480 --> 0:15:08.760 high frequency modular power electronics that we're designing here in California. 0:15:08.960 --> 0:15:11.680 You're talking about the cussing age of semiconductors in that field. 0:15:11.720 --> 0:15:14.760 We often, for example, talk about silicon carbide on this show. Yeah, 0:15:14.760 --> 0:15:16.880 but what did if you just give the backstory, did 0:15:16.920 --> 0:15:21.440 you design this technology? Are you working with someone who did? 0:15:22.640 --> 0:15:23.880 We're developing it in house. 0:15:24.640 --> 0:15:30.320 We're taking the best known topologies for advanced power electronics. 0:15:30.400 --> 0:15:32.920 There's a lot of great academic work actually in this 0:15:33.000 --> 0:15:37.840 space and applying them to the grid scale power electronics problem. Historically, 0:15:38.200 --> 0:15:40.200 the way it's been done is you have a low 0:15:40.600 --> 0:15:45.240 voltage converter inverter. Rectifier has lots of different names, going 0:15:45.280 --> 0:15:46.440 from somewhere around one. 0:15:46.360 --> 0:15:47.120 Thousand vold STC. 0:15:47.280 --> 0:15:50.280 That's what batteries are as a solar is data centers 0:15:50.280 --> 0:15:52.040 will be in the future. In fact, on this show 0:15:52.080 --> 0:15:54.800 and video talked about a intervolt power electronics their data 0:15:54.800 --> 0:15:59.560 center did. Yeah, and you know you'd go through a 0:15:59.600 --> 0:16:02.080 low voltage inverter so you'd end up with AC at 0:16:02.080 --> 0:16:03.800 four eighty or six hundred volts, and then you need 0:16:03.960 --> 0:16:06.560 one of these old school you know, grain orients and 0:16:06.600 --> 0:16:09.400 electric steel oil field transformers to get up to thirty 0:16:09.400 --> 0:16:11.440 four thousand volts, And we're going to do it directly 0:16:11.480 --> 0:16:12.960 without the use of a transformer. 0:16:13.080 --> 0:16:16.520 Drew, Who is it you're disrupting here at the moment? 0:16:16.680 --> 0:16:19.400 Is it old school electricity players? 0:16:19.520 --> 0:16:22.280 Is it the Schneider Electrics? Is that the zeemans? Who 0:16:22.360 --> 0:16:23.720 is it that you're really taking on here? 0:16:25.760 --> 0:16:26.040 Yeah? 0:16:26.120 --> 0:16:31.720 I think we're trying to provide a more efficient, more compact, 0:16:32.200 --> 0:16:35.880 more scalable solution. And there's actually a lot of players 0:16:35.880 --> 0:16:38.040 in this space. You've got switch gear providers, You've got 0:16:38.040 --> 0:16:42.080 lots of different transformer providers, there's a lot of honestly 0:16:42.120 --> 0:16:45.280 like no name power electronics providers and some bigger name players, 0:16:46.400 --> 0:16:49.000 and so it's a little bit of everyone, I would say, 0:16:49.520 --> 0:16:53.440 But really, when I zoom out, electricity growth is poised 0:16:53.440 --> 0:16:55.960 to triple over the next couple of decades, and that's 0:16:56.000 --> 0:16:59.800 not just due to AI but also electrication of vehicles, buildings, 0:17:00.160 --> 0:17:03.760 everything else. And so there's going to be room for 0:17:03.960 --> 0:17:05.959 many players to support this growth. 0:17:06.920 --> 0:17:09.959 Drew give us that wider context because we started off 0:17:10.000 --> 0:17:12.960 the show by talking about meta looking and locking in 0:17:13.119 --> 0:17:16.879 nuclear energy for its future AI needs. How much of 0:17:16.920 --> 0:17:19.560 a bottleneck is energy? How much of a bottleneck is 0:17:19.600 --> 0:17:21.960 just the infrastructure that you're trying to rework. How much 0:17:22.000 --> 0:17:25.159 do we blame well incidents like the Spanish blackout on 0:17:25.480 --> 0:17:27.520 not modern enough infrastructure. 0:17:29.359 --> 0:17:36.080 Yeah, the electricity sector is poised at an interesting inflection point. 0:17:36.119 --> 0:17:39.040 It's really the confluence of three major trends. First, you've 0:17:39.040 --> 0:17:42.040 got demand. For the first time in decades, there's actually 0:17:42.119 --> 0:17:44.680 robust electricity demand, and that's due to number of factors. 0:17:45.040 --> 0:17:48.480 You've got electrification of buildings, transportation, You've got AI. You 0:17:48.520 --> 0:17:50.600 actually have a lot of new manufacturing load, which is 0:17:50.600 --> 0:17:52.280 exciting to see that coming into the United States and 0:17:52.280 --> 0:17:53.240 other developed countries. 0:17:54.040 --> 0:17:55.719 On the same time, you've got supply. 0:17:55.960 --> 0:18:02.280 In fact, abundant clean renewable energy is growing at explosively. 0:18:02.359 --> 0:18:04.680 To be honest, in twenty twenty four, over five hundred 0:18:04.680 --> 0:18:07.639 gigawatts of the utility scale solar was deployed. That's over 0:18:08.040 --> 0:18:11.720 half a trillion of capital investment in solar that year alone, 0:18:11.840 --> 0:18:13.880 and that number was up thirty percent year over year. 0:18:14.200 --> 0:18:18.200 So massive increase and abundant potential supply. And then the 0:18:18.280 --> 0:18:21.760 last thing is aging infrastructure, especially in the developed economies. 0:18:22.359 --> 0:18:24.720 You know, over half of the infrastructure that's out there 0:18:24.800 --> 0:18:26.560 is over thirty years old, some of it's forty five. 0:18:27.080 --> 0:18:30.119 Question drew, So to interrupt, who is your customer if 0:18:30.119 --> 0:18:32.159 you want to fix the entire grid? Who are you 0:18:32.200 --> 0:18:32.679 selling to? 0:18:33.240 --> 0:18:36.159 Initially we're selling to energy project developers and data centers 0:18:36.119 --> 0:18:39.360 at center builders like yeah. So actually we have later 0:18:39.480 --> 0:18:44.240 letters of intent signed with Cruso, interesting leading AI data 0:18:44.240 --> 0:18:47.240 center developer and cloud services provider. And we also have 0:18:47.280 --> 0:18:49.560 a letter of intent signed with Intersect Power, one of 0:18:49.600 --> 0:18:51.760 the largest renewable energy developers. 0:18:51.320 --> 0:18:54.520 Working with Google in the US. I think they are you. 0:18:55.840 --> 0:19:00.480 I'm going to ask you how you view test energy 0:19:00.480 --> 0:19:04.680 business right in the context of Heroin you were an alumni, 0:19:04.760 --> 0:19:07.159 But do you expect to work with Tesla on the 0:19:07.200 --> 0:19:10.720 kind of storage and so the side certainly possible. 0:19:10.800 --> 0:19:13.200 We haven't had discussions, but I look forward to having 0:19:13.240 --> 0:19:14.880 discussions in the future if it makes sense. 0:19:15.359 --> 0:19:17.680 But the door wouldn't be closed for any reason. 0:19:18.000 --> 0:19:19.920 That I that I could see. I mean, again, I 0:19:19.960 --> 0:19:22.560 think this is a huge problem. It's a massive sector, 0:19:22.640 --> 0:19:26.600 trillions of investment over the coming years. There's opportunity for everybody. 0:19:27.920 --> 0:19:31.640 But yeah, I think Tesla Energy has done an amazing job. 0:19:31.680 --> 0:19:33.520 I mean, obviously I put a lot of my time 0:19:33.560 --> 0:19:36.080 in there over the past decade. If you go back 0:19:36.080 --> 0:19:40.399 to twenty fourteen, you know, utility scale stage storage didn't exist, 0:19:40.480 --> 0:19:42.280 I mean, residential batteries didn't exist. 0:19:42.640 --> 0:19:43.160 And now in. 0:19:43.080 --> 0:19:46.040 Twenty twenty four over fifty gigawatts of utility scale storage 0:19:46.080 --> 0:19:48.840 was deployed around the year and in capex of like 0:19:48.880 --> 0:19:54.480 forty billion dollars. That's massive, and that year itself, I 0:19:54.480 --> 0:19:57.760 think was almost a doubling of deployment. So really at 0:19:57.800 --> 0:20:00.000 the beginning of the energy storage wave, and I expect 0:20:00.080 --> 0:20:02.640 Tesla Energy to continue to grow quickly. 0:20:02.760 --> 0:20:06.560 You were the longest serving person at Tesla, basically behind 0:20:06.600 --> 0:20:09.440 Elon for a really long time, which you just kindly 0:20:09.480 --> 0:20:12.119 share some insight in how it works at Tesla. You 0:20:12.119 --> 0:20:15.679 were one of sort of three stated executive officers. But 0:20:15.840 --> 0:20:17.520 the thing that I learned in my reporting is there's 0:20:17.520 --> 0:20:21.600 a really deep bench of talent, particularly on the energy side. 0:20:21.640 --> 0:20:24.119 At a time where everyone talks about Elon must looking 0:20:24.440 --> 0:20:25.840 elsewhere and being distracted. 0:20:26.880 --> 0:20:30.000 Yeah, I mean, I'm super proud of what you know, 0:20:30.040 --> 0:20:32.800